Intelligent monitoring method for water volume of canal based on Internet of Things

Abstract

The invention discloses a canal water volume intelligent monitoring method based on the Internet of Things, and belongs to the technical field of water volume monitoring, and the method specifically comprises the steps: setting equidistant measurement points from a water outlet of an electromechanical well to a farmland diversion port and in a canal, marking the distance between each point and the water outlet, measuring the flow of each point every unit time, and calculating the flow difference of adjacent points as index data; forming a data set by a plurality of unit time data, preprocessing, and inputting a deep learning model to extract features; based on feature clustering, detecting the number of similar data sets in the radius, if the number exceeds a threshold value, forming a category cluster, otherwise, judging that the category cluster is abnormal; traversing all data sets, extracting a category cluster center, monitoring the numerical value change of the center in real time, and if the offset is large, determining that the remarking is abnormal; measuring points corresponding to the standard category cluster center are counted as mark points, mark measuring point data are compared, and the measuring points with large difference are marked as abnormal; finally, the water channel section corresponding to the abnormal measurement point is an abnormal section, and the flow loss is checked; the abnormal flow change in the water channel can be found in time.

Description

Technical Field

[0001] The present invention relates to the technical field of water volume monitoring, and in particular to an intelligent monitoring method for water volume in a canal based on the Internet of Things. Background Art

[0002] With the continuous growth of the global population and the continuous advancement of agricultural technology, the demand for agricultural water has shown a significant growth trend. In modern agricultural irrigation systems, canals are important channels for transporting water to farmland. They not only provide a stable water source, but also achieve efficient use of water resources through advanced technical means.

[0003] However, with the continuous increase in water consumption, how to accurately and efficiently monitor the water volume in the canal to ensure the rational allocation and effective use of water resources has become an urgent problem to be solved. Therefore, strengthening the monitoring of canal water volume, especially ensuring whether the canal water volume is effectively used, is of great significance to ensure the smooth progress of agricultural production, improve the efficiency of water resource utilization, and promote the sustainable development of agriculture. The application of Internet of Things technology has become an effective solution. By installing sensors and intelligent control systems in the channel, real-time monitoring and management of water consumption can be achieved.

[0004] However, in actual applications, due to leakage and blockage of canals, part of the water flow is wasted during the transportation process, resulting in a large difference between the actual amount of water reaching the farmland and the expected amount. This difference not only directly affects the growth and yield of crops, and thus affects agricultural production efficiency, but also causes a large waste of water resources. Summary of the invention

[0005] The purpose of the present invention is to provide a method for intelligent monitoring of water volume in a canal based on the Internet of Things to solve the following technical problems:

[0006] Due to leakage and blockage in the canals, part of the water is wasted during the transportation process, resulting in a large difference between the actual amount of water reaching the farmland and the expected amount.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A method for intelligently monitoring water volume in a canal based on the Internet of Things comprises the following steps:

[0009] From the outlet of the electromechanical well to the diversion outlet of the farmland, M measuring points with a fixed interval of d are set in the canal. The canal flow rate X at the measuring point is collected based on the flow sensor every unit time. i,j , X i,j It represents the canal flow rate at the jth measuring point in the i-th unit time and is transmitted to the data collector;

[0010] Through the data logger, Xi,j -X i,j-1 The indicator data of the j-th measurement point at the i-th unit time are marked, i∈[1,N], j∈[2,M], M and N are both positive integers, and the indicator data measured at multiple unit times are marked as a data set, and the data set is preprocessed; the preprocessed data set is processed by the deep learning model preset in the central control processor, and the corresponding data features are output;

[0011] Clustering the data set based on the data features, detecting the number m of similar data sets within a preset clustering radius r of any data set, if m is greater than a preset threshold, generating a category cluster based on the data set; if m is less than a preset threshold, determining the data set as an abnormal data set, and marking the corresponding measurement point as an abnormal measurement point; traversing all data sets in turn;

[0012] Extract the cluster center of each category cluster, and calculate the real-time value of the cluster center in each feature dimension. If the difference between the real-time value of the cluster center and the initial value during the clustering process is greater than the preset threshold, the cluster center will be greatly offset, and the indicator data in the category cluster will be regarded as an abnormal data set, and the corresponding measurement point will be marked as an abnormal measurement point.

[0013] Count all the remaining data sets of the cluster centers, mark the measurement points corresponding to the data sets of the centers as landmark measurement points, obtain the canal data of the landmark measurement points, select the canal data of each landmark measurement point in turn and compare them with the remaining landmark measurement points, if there is a landmark measurement point whose difference value with the canal data of all the landmark measurement points is greater than the preset threshold, then mark the landmark measurement point as an abnormal measurement point;

[0014] The water channel section corresponding to the abnormal measurement point is marked as an abnormal section and sent to the user terminal through the wireless communication interface, prompting the user to check the water flow loss of the abnormal section.

[0015] As a further solution of the present invention: the flow sensor includes a radar flow velocity sensor, a radar water level sensor and a flow calculation module, and the flow calculation module calculates the channel flow according to the data collected by the radar flow velocity sensor and the radar water level sensor.

[0016] As a further solution of the present invention: the measuring point is not only arranged at the diversion outlet from the electromechanical well to the farmland, but also arranged between two adjacent diversion outlets of the farmland.

[0017] As a further solution of the present invention: the pretreatment process is:

[0018] Filter out the outliers in the data set that exceed the preset range, delete the outliers, fill them through interpolation, and filter the data set.

[0019] As a further solution of the present invention: the filtering process is: filtering the data set based on the weighted moving average method, and the calculation formula is:

[0020]

[0021] Where n is the effective window size, n<N, w i is the weight of the i-th unit time, and the weight increases with time. n,j It represents the weighted moving average of the indicator data outside the window, which is used to filter the indicator data outside the window, that is, to merge the indicator data outside the window into one H n,j To express.

[0022] As a further solution of the present invention: the comparison process of the difference value of the marking measurement point is:

[0023] The canal data include surface flow velocity, canal water surface width and water level. A pair of marked measurement points are selected in turn, and the canal data of one of the marked measurement points is marked as a pending data set, and the canal data of the other marked measurement point is marked as a verification data set. The surface flow velocity A, canal water surface width B and water level C in the pending data set are collected and marked as pending indicators. The water surface flow velocity a, canal water surface width b and water level c in the verification data set are collected and marked as verification indicators. The pending indicators and verification indicators are respectively converted into two dimensions, and the parameters of the two-dimensional figure are marked in sequence. The two-dimensional figure is a trapezoid, and the parameters are height, upper base and lower base, respectively. The pending two-dimensional figure and the verification two-dimensional figure are respectively obtained. The pending two-dimensional figure and the verification two-dimensional figure are compared in shape to obtain the shape similarity Sim, and the difference value is 1-Sim.

[0024] As a further solution of the present invention: the process of shape comparison is:

[0025] According to the parameters, the area S1 of the pending two-dimensional image is calculated, the area S2 of the verification two-dimensional image is calculated, the pending two-dimensional image and the verification two-dimensional image are superimposed on a plane, the superposition state when the superposition area is the largest is selected, the area S3 of the superposition area in the current state is obtained, and the value of S3 / (S1+S2) is calculated, which is the shape similarity.

[0026] As a further solution of the present invention: the preset clustering radius is r=R 2 / Z, where R is the maximum Euclidean distance between all data sets and Z is the number of all data sets.

[0027] As a further solution of the present invention: the wireless communication interface is based on a GPRS communication interface and is connected to the cloud server via a TCP / UDP protocol.

[0028] Beneficial effects of the present invention:

[0029] The present invention constructs a canal flow monitoring system based on the Internet of Things technology, monitors the canal flow at the measuring point through a sensor, and takes the flow loss between two measuring points as the index data. Ideally, the water loss of the canal due to infiltration or evaporation should be uniform, so the water flow loss at a fixed distance should also be small. Therefore, the present invention first extracts the index data of multiple measuring points and multiple unit times through deep learning features as the basis for clustering, and combines the cluster analysis technology to effectively detect abnormal data sets, thereby enhancing the ability of abnormal detection; in addition, through real-time numerical monitoring of the cluster center of the category cluster, the offset of the cluster center can be discovered in time, thereby identifying the water flow loss of the abnormal segmentation, and then by comparing and analyzing the difference values ​​of the marked measuring points, the differences between the data sets are further accurately reflected, and finally, the canal sections corresponding to all abnormal data sets are marked as abnormal segments, and prompts for inspection. This multi-level abnormal detection mechanism enhances the stability and reliability of the system and ensures the accuracy of the monitoring results. By timely discovering and handling abnormal problems in the canal, the present invention can ensure that farmland obtains sufficient irrigation water, thereby improving the growth and yield of crops. This is of great significance for ensuring the smooth progress of agricultural production and improving agricultural production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described below in conjunction with the accompanying drawings.

[0031] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] See also Figure 1 As shown, the present invention is a method for intelligent monitoring of water volume in a canal based on the Internet of Things, comprising the following steps:

[0034] 1. Data Collection:

[0035] In the canal between the outlet of the electromechanical well and the diversion outlet of the farmland, 20 measurement points with a fixed interval of 50 meters are set. The distance between each measurement point and the outlet is 50 meters, 100 meters, ..., 1000 meters, which is used for spatial positioning in subsequent data analysis. Every 10 minutes, the flow sensor equipment is used to obtain the canal flow rate X at each measurement point.i,j , where X i,j represents the canal flow rate at the jth measuring point in the i-th unit time. The canal flow rate collected by the flow sensor is processed by the data logger to calculate the flow difference between adjacent measuring points, that is, X i,j -X i,j-1 ,As indicator data, the indicator data measured in multiple unit times are marked as data sets, which will be used to reflect the changes of the flow in the canal.

[0036] 2. Dataset preprocessing:

[0037] The data collector further preprocesses the index data, including data cleaning (removing outliers and noise), normalization (scaling the data to the same range), etc., to ensure data quality and reduce noise interference. The preprocessed data will be more suitable for training and analysis of deep learning models.

[0038] 3. Feature extraction:

[0039] Choosing a suitable deep learning model is crucial for feature extraction. Commonly used models include convolutional neural networks (CNN), recurrent neural networks (RNN), long short-term memory networks (LSTM) and their variants. The specific model to choose depends on the characteristics of the data and the analysis objectives.

[0040] Model architecture design: Design a suitable model architecture based on the characteristics of canal flow data. For example, for time series data, LSTM or GRU (Gated Recurrent Unit) are better choices because they can capture temporal dependencies.

[0041] Model training: Train the model using the preprocessed dataset. This usually involves defining a loss function (such as mean squared error), choosing an optimizer (such as Adam), and setting appropriate hyperparameters (such as learning rate, batch size, etc.).

[0042] Model validation and tuning: Evaluate the performance of the model through methods such as cross-validation, and adjust hyperparameters as needed to optimize the performance of the model.

[0043] Once the model is trained, it can be used to extract useful features from the data. The model is used in the central control processor to extract features from the data set. These features include but are not limited to: Time series patterns: The model can learn the time dependencies in the data, such as periodic changes, trends, etc. Flow change trends: By analyzing flow data at consecutive time points, the model can predict future flow trends. Seasonal fluctuations: For canal flow data that is significantly affected by seasons, the model can capture the characteristics of this periodic fluctuation. Anomaly detection: The model can also learn normal patterns in the data, making it easier to identify outliers or abnormal patterns that deviate from the normal range.

[0044] 4. Cluster analysis:

[0045] First, set a clustering radius r, which is an important parameter for judging the similarity between data sets. The size of the radius r should be reasonably set according to the specific application scenario and data characteristics. For each data set, calculate its similarity with other data sets within the preset clustering radius r (such as through the distance measurement in the feature space). Count the number of similar data sets m. If the number of similar data sets m of a data set is greater than the preset threshold, it is considered that these data sets have a high similarity, and a new category cluster can be generated based on the data set. This category cluster represents a typical pattern or state of canal flow data. On the contrary, if m is less than the preset threshold, it means that the data set is quite different from most other data sets, so it is judged as an abnormal data set, and its corresponding measurement point is marked as an abnormal measurement point. This helps to timely discover potential problems or abnormal conditions in the canal system. Traverse all data sets in turn and repeat the above process until all data sets have been processed. In this way, multiple category clusters and a list of abnormal data sets can be obtained.

[0046] 5. Category cluster center offset detection:

[0047] For each generated category cluster, extract its cluster center. The cluster center is the average feature representation of all data sets within the category cluster, reflecting the typical characteristics of the category cluster. During system operation, the real-time values ​​of the cluster center of each category cluster in each feature dimension are continuously calculated. This helps to monitor whether the characteristics of the category cluster change over time. If, during the clustering process, it is found that the difference between the cluster center of a category cluster and the initial value is greater than the preset threshold, it indicates that the characteristics of the category cluster have shifted significantly. This may be due to internal changes in the canal system or external interference. Once a large shift in the cluster center is detected, all indicator data within the category cluster are re-determined as abnormal data sets, and the corresponding measurement points are marked as new abnormal measurement points. This helps to capture and handle new abnormal situations in the canal system in a timely manner.

[0048] 6. Comparison of marked measurement points:

[0049] The data sets of all the remaining cluster centers are counted, and the measurement points corresponding to the data sets in the center are marked as landmark measurement points. The canal data of the landmark measurement points are obtained, and the canal data of each landmark measurement point is selected in turn for comparison with the remaining landmark measurement points. If there is a landmark measurement point whose canal data difference value with all the landmark measurement points is greater than the preset threshold, the landmark measurement point is marked as an abnormal measurement point. This comparative analysis helps to further screen and confirm abnormal data sets and improve the accuracy of anomaly detection.

[0050] 7. Abnormal segmentation prompt:

[0051] According to the results of the above steps, the canal section corresponding to the abnormal measurement point is marked as an abnormal segment. The data is transmitted to the monitoring center or user terminal through wireless communication, and a prompt is issued to the relevant personnel, suggesting that the water flow loss of the abnormal segment be inspected and maintained. This includes on-site inspection, repairing damaged parts, adjusting canal design and other measures to ensure the normal operation of the irrigation system and the effective use of water resources, and uploading the data to the cloud server for users to view online.

[0052] In a preferred embodiment of the present invention, the flow sensor system is a comprehensive solution that integrates advanced sensing technology and intelligent data processing, aiming to achieve accurate monitoring and management of canal flow. The system is mainly composed of three core components: radar flow velocity sensor, radar water level sensor and flow calculation module.

[0053] The radar flow sensor is the key component in the system responsible for directly measuring the water flow velocity. It uses high-frequency electromagnetic waves (usually millimeter wave band) to transmit and receive signals reflected from the surface of the water flow. By analyzing the frequency changes of these reflected signals, namely the Doppler shift phenomenon, the radar flow sensor can determine the speed of the water flow non-contactly. This measurement method is not only fast and accurate, but also not affected by water quality, color or transparency, and can work stably even in severe weather conditions.

[0054] Radar water level sensors focus on monitoring the water level in the canal. Also based on radar technology, the sensor transmits microwave signals to the water surface and receives its echo, thereby calculating the precise distance between the water surface and the sensor. This data is crucial for determining the actual water flow area of ​​the canal, because the flow rate is the product of the flow rate and the water flow area. The high sensitivity and anti-interference ability of the radar water level sensor ensure that reliable water level readings can be obtained even in water bodies with large fluctuations or containing impurities.

[0055] The flow calculation module receives real-time data from the radar flow velocity sensor and the radar water level sensor and processes it using advanced algorithms. First, based on the data provided by the radar water level sensor, the flow calculation module calculates the current water flow area. Then, combined with the flow velocity information provided by the radar flow velocity sensor, the instantaneous flow is calculated using a specific mathematical model (such as the average flow velocity method, flow velocity distribution integral method, etc.).

[0056] In another preferred embodiment, in order to more comprehensively monitor the operation status of the entire irrigation system, the measurement points are not only set on the water channel from the electromechanical well to the farmland diversion outlet, but also additionally set on the water channel section between two adjacent farmland diversion outlets. Such a layout can capture the changes in water channel flow in each section in more detail, which helps to discover local anomalies such as leakage and blockage, so as to take timely repair measures to ensure the efficient operation of the entire irrigation system.

[0057] In another preferred embodiment of the present invention, the pretreatment process is:

[0058] First, the collected raw data set is thoroughly reviewed to identify outliers that fall outside the preset range. These outliers may be caused by sensor failure, external interference, or errors in the data collection process. Once such outliers are detected, the system will automatically remove them from the data set.

[0059] After removing outliers, there may be gaps or missing data points left in the data set. In order to maintain the continuity and integrity of the data, advanced interpolation algorithms are used to fill in these missing data points. The choice of interpolation method depends on the characteristics of the data and the required accuracy. Common methods include linear interpolation, polynomial interpolation, or more complex machine learning algorithms.

[0060] After the interpolation process, the entire data set is further filtered to remove any possible remaining noise or irregular fluctuations.

[0061] In a preferred case of this embodiment, the data filtering process uses a weighted moving average method to process the time series data of the data set. This method is particularly suitable for processing time series data, and can focus on recent fluctuations in the data set while retaining long-term trends. The calculation formula is:

[0062]

[0063] Among them, n is the size of the effective window, n<N, that is, only the indicator data in the recent window n in the data set is retained as valid data, w i is the weight of the i-th unit time, and the weight increases with time. n,jIt represents the weighted moving average of the indicator data outside the window, which is used to filter the indicator data outside the window, that is, to merge the indicator data outside the window into one H n,j To express, H n,j It is retained together with the indicator data in the window as the data basis for subsequent processing. According to an embodiment of the present invention, these weights are increased in chronological order, which means that the data points closer to the effective window have a greater impact on the weighted moving average, while the earlier data points have a smaller impact. This weight setting helps the model to consider the historical trends outside the window while considering the most recent data in the window, and can be set to wk=1-(k / Nn), where k is an index starting from 0. In this way, the latest data point outside the window (k=0) will have the largest weight 1, and the oldest data point outside the window (k=Nn-1) will have the smallest weight 1 / Nn.

[0064] In another preferred embodiment of the present invention, the comparison process of the difference values ​​of the marking measurement points is:

[0065] Selection and marking of landmark measurement points:

[0066] This step is the starting point of the entire comparison process. The key lies in how to reasonably select and mark the landmark measurement points involved in the comparison. First, from all the existing landmark measurement point data sets, a pair of measurement points are selected in a certain order or based on specific selection rules (such as random sampling) for comparison. One point is designated as the "pending data set" and the other as the "verification data set." This pairing method ensures the comprehensiveness and fairness of the comparison and avoids the bias that may be caused by a single perspective.

[0067] Feature extraction and normalization:

[0068] After selecting the measurement points, the next step is to deeply explore the canal characteristic information of each measurement point. Specifically, three core features need to be extracted from the pending data set and the verification data set: surface flow velocity (represented as A and a, respectively), canal water surface width (represented as B and b, respectively), and water level (represented as C and c, respectively). They are marked as pending indicators and verification indicators, respectively. These characteristics are key parameters for describing the flow state of the canal and can comprehensively reflect the hydrodynamic characteristics of the canal. In order to eliminate the influence of different dimensions between different features, these original characteristic values ​​need to be standardized, that is, they need to be converted into dimensionless values ​​for subsequent comparison and analysis.

[0069] Parameterization of 2D graphics:

[0070] In order to more intuitively display and compare the characteristic differences of different measuring points, this embodiment innovatively adopts the method of two-dimensional graphic parameterization. Specifically, the pending indicators and verification indicators are cleverly converted into trapezoidal graphics on a two-dimensional plane. Each trapezoid is defined by three key parameters: height, upper base and lower base, which correspond to the converted eigenvalues ​​respectively. In this way, the data of each measuring point is given a unique "shape signature", namely a trapezoid, which contains the key water channel characteristic information of the measuring point.

[0071] Shape similarity calculation:

[0072] After converting the data into trapezoidal figures, the next task is to calculate the shape similarity between the two trapezoids (i.e., the pending two-dimensional figure and the verification two-dimensional figure). This step first involves calculating the areas S1 and S2 of the two trapezoids, which is based on the geometric properties of the trapezoids. Subsequently, by superimposing the two trapezoids on the plane and constantly adjusting their positions and angles, find the optimal state in which the superimposed area S3 reaches the maximum. Finally, the shape similarity Sim is defined as the ratio of the area of ​​the superimposed area in the current optimal state to the sum of the total areas of the two trapezoids, that is, Sim = S3 / (S1+S2). The closer this ratio is to 1, the more similar the shapes of the two trapezoids are, and vice versa, the greater the difference.

[0073] Difference value determination:

[0074] Based on the shape similarity Sim calculated above, the final difference value is given by the formula 1-Sim. This difference value is a quantitative indicator used to measure the degree of dissimilarity between two marked measurement points. Due to the characteristics of two-dimensional images, the difference in any characteristic index will be directly reflected in the shape change of the formed trapezoidal image, so this method can sensitively capture even small differences. The larger the difference value, the more significant the difference in the channel characteristics between the two measurement points, which provides strong data support for subsequent decision-making.

[0075] In another preferred embodiment of the present invention, in order to optimize the effect and efficiency of cluster analysis, we introduce a preset cluster radius parameter. This cluster radius is set based on the global characteristics of the data set. Specifically, its calculation method is to divide the maximum Euclidean distance between all data sets (denoted as R) by the total number of data sets (denoted as Z), that is, r = R 2 / Z.

[0076] In another preferred embodiment of the present invention, the wireless communication interface adopts a GPRS communication interface based on the general packet radio service technology. This choice not only ensures the stability and reliability of data transmission, but also makes full use of the existing mobile communication network infrastructure, making the deployment of the entire system more flexible and economical; the wireless communication interface can establish a connection with a remote cloud server through TCP (Transmission Control Protocol) or UDP (User Datagram Protocol), and with the help of the powerful capabilities of cloud computing, the system scale can be easily expanded to support more device access.

[0077] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for intelligent monitoring of water volume in a canal based on the Internet of Things, characterized in that: The following steps are involved: From the outlet of the electromechanical well to the diversion outlet of the farmland, M measuring points with a fixed interval of d are set in the canal. The canal flow rate X at the measuring point is collected based on the flow sensor every unit time. i,j , X i,j It represents the canal flow rate at the jth measuring point in the i-th unit time and is transmitted to the data collector; Through the data logger, X i,j -X i,j-1 The indicator data of the j-th measurement point at the i-th unit time are marked, i∈[1,N], j∈[2,M], M and N are both positive integers, and the indicator data measured at multiple unit times are marked as a data set, and the data set is preprocessed; the preprocessed data set is processed by the deep learning model preset in the central control processor, and the corresponding data features are output; Clustering the data set based on the data features, detecting the number m of similar data sets within a preset clustering radius r of any data set, and if m is greater than a preset threshold, generating a category cluster based on the data set; If m is less than the preset threshold, the data set is judged as an abnormal data set, and the corresponding measurement point is marked as an abnormal measurement point; Traverse all data sets in turn; Extract the cluster center of each category cluster, and calculate the real-time value of the cluster center in each feature dimension. If the difference between the real-time value of the cluster center and the initial value during the clustering process is greater than the preset threshold, the cluster center will be greatly offset, and the indicator data in the category cluster will be regarded as an abnormal data set, and the corresponding measurement point will be marked as an abnormal measurement point. Count all the remaining data sets of the cluster centers, mark the measurement points corresponding to the data sets of the centers as landmark measurement points, obtain the canal data of the landmark measurement points, select the canal data of each landmark measurement point in turn and compare them with the remaining landmark measurement points, if there is a landmark measurement point whose difference value with the canal data of all the landmark measurement points is greater than the preset threshold, then mark the landmark measurement point as an abnormal measurement point; The water channel section corresponding to the abnormal measurement point is marked as an abnormal section and sent to the user terminal through the wireless communication interface, prompting the user to check the water flow loss of the abnormal section.

2. According to the method of intelligent monitoring of water volume in a canal based on the Internet of Things in claim 1, it is characterized in that: The flow sensor comprises a radar flow velocity sensor, a radar water level sensor and a flow calculation module. The flow calculation module calculates the channel flow according to data collected by the radar flow velocity sensor and the radar water level sensor.

3. According to the method of intelligent monitoring of water volume in a canal based on the Internet of Things in claim 1, it is characterized in that: The measuring point is not only arranged at the diversion outlet from the electromechanical well to the farmland, but also arranged between two adjacent diversion outlets of the farmland.

4. According to the method of intelligent monitoring of water volume in a canal based on the Internet of Things in claim 1, it is characterized in that: The pre-processing process is as follows: Filter out the outliers in the data set that exceed the preset range, delete the outliers, fill them through interpolation, and filter the data set.

5. The method for intelligent monitoring of water volume in a canal based on the Internet of Things according to claim 4 is characterized in that: The filtering process is: filtering the data set based on the weighted moving average method, and the calculation formula is: Where n is the effective window size, n<N, w i is the weight of the i-th unit time, and the weight increases with time. n,j It represents the weighted moving average of the indicator data outside the window, which is used to filter the indicator data outside the window, that is, to merge the indicator data outside the window into one H n,j To express.

6. The method for intelligent monitoring of water volume in a canal based on the Internet of Things according to claim 1 is characterized in that: The comparison process of the difference value of the mark measurement point is: The canal data include surface flow velocity, canal water surface width and water level. A pair of marked measurement points are selected in turn, and the canal data of one of the marked measurement points is marked as a pending data set, and the canal data of the other marked measurement point is marked as a verification data set. The surface flow velocity A, canal water surface width B and water level C in the pending data set are collected and marked as pending indicators. The water surface flow velocity a, canal water surface width b and water level c in the verification data set are collected and marked as verification indicators. The pending indicators and verification indicators are respectively converted into two dimensions, and the parameters of the two-dimensional figure are marked in sequence. The two-dimensional figure is a trapezoid, and the parameters are height, upper base and lower base, respectively. The pending two-dimensional figure and the verification two-dimensional figure are respectively obtained. The pending two-dimensional figure and the verification two-dimensional figure are compared in shape to obtain the shape similarity Sim, and the difference value is 1-Sim.

7. The method for intelligent monitoring of water volume in a canal based on the Internet of Things according to claim 6 is characterized in that: The process of shape comparison is: According to the parameters, the area S1 of the pending two-dimensional image is calculated, the area S2 of the verification two-dimensional image is calculated, the pending two-dimensional image and the verification two-dimensional image are superimposed on a plane, the superposition state when the superposition area is the largest is selected, the area S3 of the superposition area in the current state is obtained, and the value of S3 / (S1+S2) is calculated, which is the shape similarity.

8. The method for intelligent monitoring of water volume in a canal based on the Internet of Things according to claim 1 is characterized in that: The preset clustering radius is r=R 2 / Z, where R is the maximum Euclidean distance between all data sets and Z is the number of all data sets.

9. The method for intelligent monitoring of water volume in a canal based on the Internet of Things according to claim 1, characterized in that: The wireless communication interface is based on the GPRS communication interface and is connected to the cloud server via the TCP / UDP protocol.

Images